import os
import gradio as gr
import cv2
import numpy as np
import mediapipe as mp
from sklearn.linear_model import LinearRegression
import random
import base64
import joblib
from datetime import datetime
import shutil
from reportlab.lib.pagesizes import letter
from reportlab.platypus import SimpleDocTemplate, Paragraph, Spacer, Table, TableStyle
from reportlab.lib.styles import getSampleStyleSheet, ParagraphStyle
from reportlab.lib import colors
import atexit
import glob
# Cleanup temporary files on exit
def cleanup_temp_files():
for temp_file in glob.glob("/tmp/Health_Report_*.pdf"):
os.remove(temp_file)
atexit.register(cleanup_temp_files)
# Initialize the face mesh model
mp_face_mesh = mp.solutions.face_mesh
face_mesh = mp_face_mesh.FaceMesh(static_image_mode=True,
max_num_faces=1,
refine_landmarks=True,
min_detection_confidence=0.5)
# Functions for feature extraction
def extract_features(image, landmarks):
red_channel = image[:, :, 2]
green_channel = image[:, :, 1]
blue_channel = image[:, :, 0]
red_percent = 100 * np.mean(red_channel) / 255
green_percent = 100 * np.mean(green_channel) / 255
blue_percent = 100 * np.mean(blue_channel) / 255
return [red_percent, green_percent, blue_percent]
def train_model(output_range):
X = [[
random.uniform(0.2, 0.5),
random.uniform(0.05, 0.2),
random.uniform(0.05, 0.2),
random.uniform(0.2, 0.5),
random.uniform(0.2, 0.5),
random.uniform(0.2, 0.5),
random.uniform(0.2, 0.5)
] for _ in range(100)]
y = [random.uniform(*output_range) for _ in X]
model = LinearRegression().fit(X, y)
return model
# Load models
try:
hemoglobin_model = joblib.load("hemoglobin_model_from_anemia_dataset.pkl")
spo2_model = joblib.load("spo2_model_simulated.pkl")
hr_model = joblib.load("heart_rate_model.pkl")
except FileNotFoundError:
print("Error: One or more .pkl model files are missing. Please upload them.")
exit(1)
models = {
"Hemoglobin": hemoglobin_model,
"WBC Count": train_model((4.0, 11.0)),
"Platelet Count": train_model((150, 450)),
"Iron": train_model((60, 170)),
"Ferritin": train_model((30, 300)),
"TIBC": train_model((250, 400)),
"Bilirubin": train_model((0.3, 1.2)),
"Creatinine": train_model((0.6, 1.2)),
"Urea": train_model((7, 20)),
"Sodium": train_model((135, 145)),
"Potassium": train_model((3.5, 5.1)),
"TSH": train_model((0.4, 4.0)),
"Cortisol": train_model((5, 25)),
"FBS": train_model((70, 110)),
"HbA1c": train_model((4.0, 5.7)),
"Albumin": train_model((3.5, 5.5)),
"BP Systolic": train_model((90, 120)),
"BP Diastolic": train_model((60, 80)),
"Temperature": train_model((97, 99))
}
# Helper function for risk level color coding
def get_risk_color(value, normal_range):
low, high = normal_range
if value < low:
return ("Low", "๐ป", "#fff3cd")
elif value > high:
return ("High", "๐บ", "#f8d7da")
else:
return ("Normal", "โ
", "#d4edda")
# Function to build table for test results
def build_table(title, rows):
# Define color mapping
color_map = {
"Normal": "#28a745",
"High": "#dc3545",
"Low": "#ffc107"
}
html = (
f''
f'
'
f'
{title}
'
f''
f'
'
f'| Test | Result | Range | Level |
|---|
'
)
for i, (label, value, ref) in enumerate(rows):
level, icon, bg = get_risk_color(value, ref)
row_bg = "#f8f9fa" if i % 2 == 0 else "white"
if level != "Normal":
row_bg = bg
if "Count" in label or "Platelet" in label:
value_str = f"{value:.0f}"
else:
value_str = f"{value:.2f}"
html += f'| {label} | {value_str} | {ref[0]} - {ref[1]} | {icon} {level} |
'
html += '
HEALTH CARD
Report Date: {current_date}
{f'
Patient ID: {patient_id}
' if patient_id else ''}
{patient_name if patient_name else 'Lab Test Results'}
{f'Age: {patient_age} | Gender: {patient_gender}' if patient_age and patient_gender else 'AI-Generated Health Analysis'}
Face-Based Health Analysis Report
{test_results['Hematology']}
{test_results['Iron Panel']}
{test_results['Liver & Kidney']}
{test_results['Electrolytes']}
{test_results['Vitals']}
๐ Summary & Recommendations
{summary}
"""
return html
# Initialize global variable for patient details
current_patient_details = {'name': '', 'age': '', 'gender': '', 'id': ''}
# Modified analyze_face function
def analyze_face(input_data):
if isinstance(input_data, str): # Video input (file path in Replit)
cap = cv2.VideoCapture(input_data)
if not cap.isOpened():
return "โ ๏ธ Error: Could not open video.
", None
ret, frame = cap.read()
cap.release()
if not ret:
return "โ ๏ธ Error: Could not read video frame.
", None
else: # Image input
frame = input_data
if frame is None:
return "โ ๏ธ Error: No image provided.
", None
# Resize image to reduce processing time
frame = cv2.resize(frame, (640, 480))
frame_rgb = cv2.cvtColor(frame, cv2.COLOR_BGR2RGB)
result = face_mesh.process(frame_rgb)
if not result.multi_face_landmarks:
return "โ ๏ธ Error: Face not detected.
", None
landmarks = result.multi_face_landmarks[0].landmark
features = extract_features(frame_rgb, landmarks)
test_values = {}
for label in models:
if label == "Hemoglobin":
prediction = models[label].predict([features])[0]
test_values[label] = prediction
else:
value = models[label].predict([[random.uniform(0.2, 0.5) for _ in range(7)]])[0]
test_values[label] = value
gray = cv2.cvtColor(frame_rgb, cv2.COLOR_RGB2GRAY)
green_std = np.std(frame_rgb[:, :, 1]) / 255
brightness_std = np.std(gray) / 255
tone_index = np.mean(frame_rgb[100:150, 100:150]) / 255 if frame_rgb[100:150, 100:150].size else 0.5
hr_features = [brightness_std, green_std, tone_index]
heart_rate = float(np.clip(hr_model.predict([hr_features])[0], 60, 100))
skin_patch = frame_rgb[100:150, 100:150]
skin_tone_index = np.mean(skin_patch) / 255 if skin_patch.size else 0.5
brightness_variation = np.std(cv2.cvtColor(frame_rgb, cv2.COLOR_RGB2GRAY)) / 255
spo2_features = [heart_rate, brightness_variation, skin_tone_index]
spo2 = spo2_model.predict([spo2_features])[0]
rr = int(12 + abs(heart_rate % 5 - 2))
test_results = {
"Hematology": build_table("๐ฉธ Hematology", [("Hemoglobin", test_values["Hemoglobin"], (13.5, 17.5)),
("WBC Count", test_values["WBC Count"], (4.0, 11.0)),
("Platelet Count", test_values["Platelet Count"], (150, 450))]),
"Iron Panel": build_table("๐งฌ Iron Panel", [("Iron", test_values["Iron"], (60, 170)),
("Ferritin", test_values["Ferritin"], (30, 300)),
("TIBC", test_values["TIBC"], (250, 400))]),
"Liver & Kidney": build_table("๐งฌ Liver & Kidney", [("Bilirubin", test_values["Bilirubin"], (0.3, 1.2)),
("Creatinine", test_values["Creatinine"], (0.6, 1.2)),
("Urea", test_values["Urea"], (7, 20))]),
"Electrolytes": build_table("๐งช Electrolytes", [("Sodium", test_values["Sodium"], (135, 145)),
("Potassium", test_values["Potassium"], (3.5, 5.1))]),
"Vitals": build_table("โค๏ธ Vitals", [("SpO2", spo2, (95, 100)),
("Heart Rate", heart_rate, (60, 100)),
("Respiratory Rate", rr, (12, 20)),
("Temperature", test_values["Temperature"], (97, 99)),
("BP Systolic", test_values["BP Systolic"], (90, 120)),
("BP Diastolic", test_values["BP Diastolic"], (60, 80))])
}
summary = "- Your hemoglobin is a bit low โ this could mean mild anemia.
- Low iron storage detected โ consider an iron profile test.
- Elevated bilirubin โ possible jaundice. Recommend LFT.
- High HbA1c โ prediabetes indication. Recommend glucose check.
- Low SpOโ โ suggest retesting with a pulse oximeter.
"
_, buffer = cv2.imencode('.png', frame_rgb)
profile_image_base64 = base64.b64encode(buffer).decode('utf-8')
pdf_filename = f"Health_Report_{datetime.now().strftime('%Y-%m-%d_%H-%M-%S')}.pdf"
pdf_result, pdf_filepath = save_results_to_pdf(test_results, pdf_filename)
if pdf_filepath:
temp_pdf_path = "/tmp/" + os.path.basename(pdf_filepath)
shutil.copy(pdf_filepath, temp_pdf_path)
if os.path.exists(temp_pdf_path) and os.access(temp_pdf_path, os.R_OK):
health_card_html = build_health_card(
profile_image_base64,
test_results,
summary,
temp_pdf_path,
current_patient_details['name'],
current_patient_details['age'],
current_patient_details['gender'],
current_patient_details['id']
)
return health_card_html, temp_pdf_path
else:
return "โ ๏ธ Error: PDF file not accessible.
", None
return "โ ๏ธ Error: Failed to generate PDF.
", None
# Modified route_inputs function
def route_inputs(mode, image, video, patient_name, patient_age, patient_gender, patient_id):
if mode == "Image" and image is None:
return "โ ๏ธ Error: No image provided.
", None
if mode == "Video" and video is None:
return "โ ๏ธ Error: No video provided.
", None
global current_patient_details
current_patient_details = {
'name': patient_name,
'age': patient_age,
'gender': patient_gender,
'id': patient_id
}
health_card_html, pdf_file_path = analyze_face(image if mode == "Image" else video)
return health_card_html, pdf_file_path
# Create Gradio interface
with gr.Blocks() as demo:
gr.Markdown("""# ๐ง Face-Based Lab Test AI Report (Video Mode)""")
with gr.Row():
with gr.Column():
gr.Markdown("### Patient Information")
patient_name = gr.Textbox(label="Patient Name", placeholder="Enter patient name")
patient_age = gr.Number(label="Age", value=25, minimum=1, maximum=120)
patient_gender = gr.Radio(label="Gender", choices=["Male", "Female", "Other"], value="Male")
patient_id = gr.Textbox(label="Patient ID", placeholder="Enter patient ID (optional)")
gr.Markdown("### Image/Video Input")
mode_selector = gr.Radio(label="Choose Input Mode",
choices=["Image", "Video"],
value="Image")
image_input = gr.Image(type="numpy", label="๐ธ Upload Face Image")
video_input = gr.Video(label="Upload Face Video",
sources=["upload", "webcam"])
submit_btn = gr.Button("๐ Analyze")
with gr.Column():
result_html = gr.HTML(label="๐งช Health Report Table")
result_pdf = gr.File(label="Download Health Report PDF", interactive=False)
submit_btn.click(fn=route_inputs,
inputs=[mode_selector, image_input, video_input, patient_name, patient_age, patient_gender, patient_id],
outputs=[result_html, result_pdf])
# Launch Gradio for Replit
demo.launch(server_name="0.0.0.0", server_port=7860)