diff --git a/backend/backend/ecg_app/Ai/difficulty_clustering.py b/backend/backend/ecg_app/Ai/difficulty_clustering.py
new file mode 100644
index 0000000..8afea62
--- /dev/null
+++ b/backend/backend/ecg_app/Ai/difficulty_clustering.py
@@ -0,0 +1,75 @@
+from django.db.models import Count, Avg, Q
+from ecg_app.models import EcgSamples, QuestionAttempt
+from sklearn.cluster import KMeans
+import pandas as pd
+
+
+def recalculate_difficulty(n_clusters=5, min_attempts=5):
+    """
+    Recalculates difficulty level for each ECG sample using KMeans clustering.
+    Difficulty is based on:
+    - Error rate (1 - accuracy)
+    - Average response time
+    Only samples with at least `min_attempts` attempts are included.
+    """
+
+    print("📊 Extracting features for ECG samples...")
+
+    # Aggregate performance data per ECG sample via QuestionAttempt
+    samples = (
+        QuestionAttempt.objects.values('question__ecg_sample')
+        .annotate(
+            total=Count('id'),
+            correct=Count('id', filter=Q(is_correct=True)),
+            avg_time=Avg('response_time')
+        )
+        .filter(total__gte=min_attempts)
+    )
+
+    if not samples:
+        print("⚠️ Not enough data to run KMeans.")
+        return
+
+    data = []
+    ids = []
+
+    for s in samples:
+        accuracy = s['correct'] / s['total']
+        time = s['avg_time'] or 0
+        sample_id = s['question__ecg_sample']
+        ids.append(sample_id)
+        data.append([1 - accuracy, time])  # Features: [error_rate, avg_time]
+
+    features = pd.DataFrame(data, columns=['error_rate', 'avg_time'])
+
+    print(f"🔍 Clustering {len(ids)} samples using KMeans...")
+
+    # Apply KMeans clustering
+    model = KMeans(n_clusters=n_clusters, random_state=42)
+    clusters = model.fit_predict(features)
+    
+    # Rank clusters by difficulty (error_rate + avg_time)
+    features['cluster'] = clusters
+    features['sample_id'] = ids
+    cluster_difficulty = (
+        features.groupby('cluster')[['error_rate', 'avg_time']]
+        .mean()
+        .assign(difficulty_score=lambda df: df['error_rate'] + df['avg_time'])
+        .sort_values('difficulty_score')
+    )
+
+    cluster_to_difficulty_level = {
+        cluster_id: rank for rank, cluster_id in enumerate(cluster_difficulty.index)
+    }
+    # Save difficulty level to the database
+    updated = 0
+    for i, sample_id in enumerate(ids):
+        try:
+            sample = EcgSamples.objects.get(sample_id=sample_id)  # use sample_id (custom PK)
+            sample.difficulty_level = cluster_to_difficulty_level[clusters[i]]
+            sample.save()
+            updated += 1
+        except EcgSamples.DoesNotExist:
+            continue
+
+    print(f"✅ Difficulty levels updated for {updated} samples.")
diff --git a/backend/backend/ecg_app/Ai/student_level_estimation.py b/backend/backend/ecg_app/Ai/student_level_estimation.py
new file mode 100644
index 0000000..99e8732
--- /dev/null
+++ b/backend/backend/ecg_app/Ai/student_level_estimation.py
@@ -0,0 +1,55 @@
+from sklearn.ensemble import RandomForestClassifier
+from sklearn.preprocessing import StandardScaler
+from sklearn.pipeline import Pipeline
+import numpy as np
+from django.contrib.auth.models import User
+from ecg_app.models import QuizAttempt, QuestionAttempt, Profile, EcgDocLabels, StudentTagMastery
+
+def recalculate_student_levels():
+    print("🧠 Running student level estimation...")
+
+    # Step 1: Collect training data
+    X_train, y_train = [], []
+    user_features = {}
+    
+    for user in User.objects.all():
+        attempts = QuestionAttempt.objects.filter(quiz_attempt__user=user)
+        if not attempts.exists():
+            continue
+
+        avg_response_time = attempts.aggregate(avg_time=models.Avg('response_time'))['avg_time'] or 0
+        correct_ratio = attempts.aggregate(correct_avg=models.Avg(models.Case(
+            models.When(is_correct=True, then=1),
+            default=0,
+            output_field=models.FloatField()
+        )))['correct_avg'] or 0
+
+        total_attempts = attempts.count()
+
+        # For training: check if this user already has a labeled level
+        profile = getattr(user, 'profile', None)
+        if profile and profile.student_level is not None:
+            X_train.append([avg_response_time, correct_ratio, total_attempts])
+            y_train.append(profile.student_level)
+        
+        user_features[user.id] = [avg_response_time, correct_ratio, total_attempts]
+
+    if not X_train:
+        print("⚠️ No training data available.")
+        return
+
+    # Step 2: Train model
+    model = Pipeline([
+        ('scaler', StandardScaler()),
+        ('classifier', RandomForestClassifier(n_estimators=100, random_state=42))
+    ])
+    model.fit(X_train, y_train)
+
+    # Step 3: Predict levels for all users
+    for user_id, features in user_features.items():
+        predicted_level = model.predict([features])[0]
+        profile = Profile.objects.get(user__id=user_id)
+        profile.student_level = predicted_level
+        profile.save()
+
+    print("✅ Student levels updated successfully.")
diff --git a/backend/backend/ecg_app/Ai/tag_mastery_estimation.py b/backend/backend/ecg_app/Ai/tag_mastery_estimation.py
new file mode 100644
index 0000000..5ee3ce2
--- /dev/null
+++ b/backend/backend/ecg_app/Ai/tag_mastery_estimation.py
@@ -0,0 +1,72 @@
+from django.db.models import Avg, Count, Q
+from ecg_app.models import QuestionAttempt, EcgSamplesDocLabels, StudentTagMastery
+from sklearn.ensemble import RandomForestRegressor
+import pandas as pd
+
+
+def recalculate_tag_mastery():
+    """
+    Calculates and updates tag-specific mastery levels for each student.
+    Uses a regression model to estimate mastery as a percentage (0-100).
+    """
+    print("📊 Gathering question attempt data for tag-based mastery...")
+
+    # Step 1: Gather relevant data
+    attempts = (
+        QuestionAttempt.objects
+        .filter(question__ecg_sample__doc_labels__isnull=False)
+        .values('quiz_attempt__user', 'question__ecg_sample__doc_labels')
+        .annotate(
+            avg_correct=Avg('is_correct'),
+            avg_time=Avg('response_time'),
+            total=Count('id')
+        )
+        .filter(total__gte=3)
+    )
+
+    if not attempts:
+        print("⚠️ Not enough data to train the model.")
+        return
+
+    # Step 2: Build dataset
+    rows = []
+    for row in attempts:
+        rows.append({
+            'user_id': row['quiz_attempt__user'],
+            'tag_id': row['question__ecg_sample__doc_labels'],
+            'avg_correct': float(row['avg_correct']),
+            'avg_time': float(row['avg_time']),
+            'total_attempts': row['total']
+        })
+
+    df = pd.DataFrame(rows)
+
+    # Normalize input values
+    df['avg_time'] = df['avg_time'].fillna(df['avg_time'].mean())
+    df['avg_correct'] = df['avg_correct'].fillna(0)
+
+    # Step 3: Train a regression model to simulate mastery percentage
+    df['mastery_percent'] = df['avg_correct'] * 100  # Simple assumption
+
+    features = df[['avg_correct', 'avg_time', 'total_attempts']]
+    target = df['mastery_percent']
+
+    model = RandomForestRegressor(random_state=42)
+    model.fit(features, target)
+    df['predicted_mastery'] = model.predict(features)
+
+    print("🤖 Model trained and predictions generated.")
+
+    # Step 4: Save mastery values to DB
+    updated = 0
+    for _, row in df.iterrows():
+        mastery_value = min(100.0, max(0.0, float(row['predicted_mastery'])))  # Ensure in 0-100 range
+
+        StudentTagMastery.objects.update_or_create(
+            user_id=row['user_id'],
+            tag_id=row['tag_id'],
+            defaults={'mastery_score': mastery_value}
+        )
+        updated += 1
+
+    print(f"✅ Updated mastery scores for {updated} student-tag pairs.")
diff --git a/backend/backend/ecg_app/management/commands/recalculate_difficulty.py b/backend/backend/ecg_app/management/commands/recalculate_difficulty.py
new file mode 100644
index 0000000..d141f2d
--- /dev/null
+++ b/backend/backend/ecg_app/management/commands/recalculate_difficulty.py
@@ -0,0 +1,8 @@
+from django.core.management.base import BaseCommand
+from ecg_app.Ai.difficulty_clustering import recalculate_difficulty
+
+class Command(BaseCommand):
+    help = 'Recalculate difficulty level for each ECG sample using KMeans clustering'
+
+    def handle(self, *args, **kwargs):
+        recalculate_difficulty()
diff --git a/backend/backend/ecg_app/management/commands/recalculate_student_levels.py b/backend/backend/ecg_app/management/commands/recalculate_student_levels.py
new file mode 100644
index 0000000..64dfb05
--- /dev/null
+++ b/backend/backend/ecg_app/management/commands/recalculate_student_levels.py
@@ -0,0 +1,13 @@
+from django.core.management.base import BaseCommand
+from ecg_app.Ai.student_level_estimation import recalculate_student_levels
+
+class Command(BaseCommand):
+    help = 'Recalculate student levels using ML model based on past quiz attempts'
+
+    def handle(self, *args, **options):
+        self.stdout.write("🔄 Running student level estimation...")
+        try:
+            recalculate_student_levels()
+            self.stdout.write(self.style.SUCCESS("✅ Student levels updated successfully."))
+        except Exception as e:
+            self.stdout.write(self.style.ERROR(f"❌ Error occurred: {str(e)}"))
diff --git a/backend/backend/ecg_app/management/commands/recalculate_tag_mastery.py b/backend/backend/ecg_app/management/commands/recalculate_tag_mastery.py
new file mode 100644
index 0000000..0d9375e
--- /dev/null
+++ b/backend/backend/ecg_app/management/commands/recalculate_tag_mastery.py
@@ -0,0 +1,10 @@
+from django.core.management.base import BaseCommand
+from ecg_app.Ai.tag_mastery_estimation import recalculate_tag_mastery  
+
+class Command(BaseCommand):
+    help = 'Recalculate student mastery per ECG tag using regression model'
+
+    def handle(self, *args, **kwargs):
+        self.stdout.write(self.style.NOTICE("🔄 Starting tag-based mastery recalculation..."))
+        recalculate_tag_mastery()
+        self.stdout.write(self.style.SUCCESS("✅ Done. Mastery values updated."))
diff --git a/backend/backend/ecg_app/migrations/0011_ecgsamples_difficulty_level.py b/backend/backend/ecg_app/migrations/0011_ecgsamples_difficulty_level.py
new file mode 100644
index 0000000..e8d052b
--- /dev/null
+++ b/backend/backend/ecg_app/migrations/0011_ecgsamples_difficulty_level.py
@@ -0,0 +1,18 @@
+# Generated by Django 5.1.3 on 2025-07-05 13:08
+
+from django.db import migrations, models
+
+
+class Migration(migrations.Migration):
+
+    dependencies = [
+        ('ecg_app', '0010_alter_ecgsamplevalidation_options_and_more'),
+    ]
+
+    operations = [
+        migrations.AddField(
+            model_name='ecgsamples',
+            name='difficulty_level',
+            field=models.IntegerField(blank=True, null=True),
+        ),
+    ]
diff --git a/backend/backend/ecg_app/migrations/0012_questionattempt_response_time.py b/backend/backend/ecg_app/migrations/0012_questionattempt_response_time.py
new file mode 100644
index 0000000..303367d
--- /dev/null
+++ b/backend/backend/ecg_app/migrations/0012_questionattempt_response_time.py
@@ -0,0 +1,18 @@
+# Generated by Django 5.2.4 on 2025-07-08 08:44
+
+from django.db import migrations, models
+
+
+class Migration(migrations.Migration):
+
+    dependencies = [
+        ('ecg_app', '0011_ecgsamples_difficulty_level'),
+    ]
+
+    operations = [
+        migrations.AddField(
+            model_name='questionattempt',
+            name='response_time',
+            field=models.FloatField(blank=True, null=True),
+        ),
+    ]
diff --git a/backend/backend/ecg_app/migrations/0013_profile_student_level.py b/backend/backend/ecg_app/migrations/0013_profile_student_level.py
new file mode 100644
index 0000000..665d52b
--- /dev/null
+++ b/backend/backend/ecg_app/migrations/0013_profile_student_level.py
@@ -0,0 +1,18 @@
+# Generated by Django 5.2.4 on 2025-07-17 18:42
+
+from django.db import migrations, models
+
+
+class Migration(migrations.Migration):
+
+    dependencies = [
+        ('ecg_app', '0012_questionattempt_response_time'),
+    ]
+
+    operations = [
+        migrations.AddField(
+            model_name='profile',
+            name='student_level',
+            field=models.IntegerField(blank=True, null=True),
+        ),
+    ]
diff --git a/backend/backend/ecg_app/migrations/0014_studenttagmastery.py b/backend/backend/ecg_app/migrations/0014_studenttagmastery.py
new file mode 100644
index 0000000..32148dc
--- /dev/null
+++ b/backend/backend/ecg_app/migrations/0014_studenttagmastery.py
@@ -0,0 +1,28 @@
+# Generated by Django 5.2.4 on 2025-07-21 12:13
+
+import django.db.models.deletion
+from django.conf import settings
+from django.db import migrations, models
+
+
+class Migration(migrations.Migration):
+
+    dependencies = [
+        ('ecg_app', '0013_profile_student_level'),
+        migrations.swappable_dependency(settings.AUTH_USER_MODEL),
+    ]
+
+    operations = [
+        migrations.CreateModel(
+            name='StudentTagMastery',
+            fields=[
+                ('id', models.BigAutoField(auto_created=True, primary_key=True, serialize=False, verbose_name='ID')),
+                ('mastery_score', models.FloatField(blank=True, null=True)),
+                ('tag', models.ForeignKey(on_delete=django.db.models.deletion.CASCADE, related_name='student_masteries', to='ecg_app.ecgdoclabels')),
+                ('user', models.ForeignKey(on_delete=django.db.models.deletion.CASCADE, related_name='tag_masteries', to=settings.AUTH_USER_MODEL)),
+            ],
+            options={
+                'unique_together': {('user', 'tag')},
+            },
+        ),
+    ]
diff --git a/backend/backend/ecg_app/models.py b/backend/backend/ecg_app/models.py
index 6461c7f..7b9c359 100644
--- a/backend/backend/ecg_app/models.py
+++ b/backend/backend/ecg_app/models.py
@@ -15,6 +15,7 @@ class EcgSamples(models.Model):
         null=True
     )
     age = models.PositiveIntegerField(blank=True, null=True)
+    difficulty_level = models.IntegerField(null=True, blank=True)
 
 
     def __str__(self):
@@ -108,6 +109,7 @@ class Profile(models.Model):
         choices=[('student', 'Student'), ('teacher', 'Teacher')],
         default='student'
     )
+    student_level = models.IntegerField(null=True, blank=True)
 
     def __str__(self):
         return self.user.username
@@ -177,6 +179,7 @@ class QuestionAttempt(models.Model):
     question = models.ForeignKey(Question, on_delete=models.CASCADE)
     selected_choice = models.ForeignKey(Choice, on_delete=models.SET_NULL, blank=True, null=True)
     is_correct = models.BooleanField(default=False)
+    response_time = models.FloatField(null=True, blank=True)
 
     def __str__(self):
         return f"Attempt for Question {self.question.id} in {self.quiz_attempt}"
@@ -215,3 +218,14 @@ class Meta:
 
     def __str__(self):
         return f"{self.student.username} - {self.group.name} ({self.status})"
+
+class StudentTagMastery(models.Model):
+    user = models.ForeignKey(User, on_delete=models.CASCADE, related_name='tag_masteries')
+    tag = models.ForeignKey(EcgDocLabels, on_delete=models.CASCADE, related_name='student_masteries')
+    mastery_score = models.FloatField(null=True, blank=True)
+
+    class Meta:
+        unique_together = ('user', 'tag')
+
+    def __str__(self):
+        return f"{self.user.username} - {self.tag.label_desc} - Score: {self.mastery_score}"
diff --git a/backend/backend/ecg_app/views/quiz.py b/backend/backend/ecg_app/views/quiz.py
index 418ae52..60175e7 100644
--- a/backend/backend/ecg_app/views/quiz.py
+++ b/backend/backend/ecg_app/views/quiz.py
@@ -142,6 +142,7 @@ def create(self, request, *args, **kwargs):
         for answer in answers:
             question_id = answer.get('question')
             choice_id = answer.get('selected_choice')
+            response_time = answer.get('response_time')
 
             try:
                 question = Question.objects.get(id=question_id, quiz=quiz)
@@ -155,7 +156,8 @@ def create(self, request, *args, **kwargs):
                     quiz_attempt=quiz_attempt,
                     question=question,
                     selected_choice=choice,
-                    is_correct=is_correct
+                    is_correct=is_correct,
+                    response_time=response_time
                 )
             except (Question.DoesNotExist, Choice.DoesNotExist):
                 continue
diff --git a/backend/backend/ecg_backend/settings.py b/backend/backend/ecg_backend/settings.py
index 71cd77c..f56616b 100644
--- a/backend/backend/ecg_backend/settings.py
+++ b/backend/backend/ecg_backend/settings.py
@@ -24,7 +24,8 @@
 SECRET_KEY = os.getenv('DJANGO_SECRET_KEY')
 
 # SECURITY WARNING: don't run with debug turned on in production!
-DEBUG = os.getenv('DEBUG').lower() == 'true'
+DEBUG = os.getenv('DEBUG', 'false').lower() == 'true'
+
 PROD = not DEBUG
 
 if PROD:
diff --git a/backend/requirements.txt b/backend/requirements.txt
index 9b156fc..7514377 100644
--- a/backend/requirements.txt
+++ b/backend/requirements.txt
@@ -5,4 +5,6 @@ django-cors-headers
 django-simple-history
 tqdm
 Pillow
-gunicorn
\ No newline at end of file
+gunicorn
+pandas
+scikit-learn
\ No newline at end of file
diff --git a/frontend/src/components/Quiz.jsx b/frontend/src/components/Quiz.jsx
index 7bf6565..98918ee 100644
--- a/frontend/src/components/Quiz.jsx
+++ b/frontend/src/components/Quiz.jsx
@@ -10,6 +10,8 @@ const Quiz = () => {
   const [quizzes, setQuizzes] = useState([]);
   const [selectedQuiz, setSelectedQuiz] = useState(null);
   const [currentQuestionIndex, setCurrentQuestionIndex] = useState(0);
+  const [questionStartTime, setQuestionStartTime] = useState(Date.now());
+  const [responseTimes, setResponseTimes] = useState({});
   const [answers, setAnswers] = useState({});
   const [isLoading, setIsLoading] = useState(false);
   const [error, setError] = useState(null);
@@ -105,6 +107,7 @@ const Quiz = () => {
       setIsAnswerChecked(false);
       setCorrectChoiceId(null);
       setStartTime(Date.now());
+      setQuestionStartTime(Date.now());
       setDuration(null);
       if (quiz.questions[0]) {
         const sortedChoices = [...quiz.questions[0].choices].sort((a, b) => a.text.localeCompare(b.text));
@@ -148,6 +151,7 @@ const Quiz = () => {
     setIsAnswerChecked(false);
     setCorrectChoiceId(null);
     setStartTime(Date.now());
+    setQuestionStartTime(Date.now());
     setDuration(null);
     if (quiz.questions[0]) {
       const sortedChoices = [...quiz.questions[0].choices].sort((a, b) => a.text.localeCompare(b.text));
@@ -170,8 +174,13 @@ const Quiz = () => {
     try {
       const currentQuestion = selectedQuiz.questions[currentQuestionIndex];
       const result = await checkAnswer(currentQuestion.id, currentAnswer);
+      const responseTimeSeconds = Math.floor((Date.now() - questionStartTime) / 1000);
       setIsAnswerChecked(true);
       setCorrectChoiceId(result.correct_choice_id);
+      setResponseTimes(prev => ({
+      ...prev,
+      [currentQuestion.id]: responseTimeSeconds
+    }));
     } catch (err) {
       setError('Failed to check answer. Please try again.');
     }
@@ -182,6 +191,7 @@ const Quiz = () => {
     
     if (currentQuestionIndex < selectedQuiz.questions.length - 1) {
       setCurrentQuestionIndex(prev => prev + 1);
+      setQuestionStartTime(Date.now());
       setCurrentAnswer(null);
       setIsAnswerChecked(false);
       setCorrectChoiceId(null);
@@ -202,7 +212,8 @@ const Quiz = () => {
         quiz: selectedQuiz.id,
         answers: Object.entries(finalAnswers).map(([questionId, choiceId]) => ({
           question: parseInt(questionId),
-          selected_choice: choiceId
+          selected_choice: choiceId,
+          response_time: responseTimes[questionId] || null
         }))
       });