app.py

import streamlit as st
import pandas as pd
import numpy as np
from datetime import datetime, timedelta
from database import init_db, add_medication, get_all_medications
import matplotlib.pyplot as plt
from matplotlib.dates import DateFormatter
import seaborn as sns
import base64
from sklearn.model_selection import train_test_split
from sklearn.ensemble import GradientBoostingClassifier, IsolationForest

# Initialize the database
init_db()

# Set page config
st.set_page_config(
    page_title="Medication Tracker",
    page_icon="💊",
    layout="centered",
    initial_sidebar_state="expanded",
)

# Custom CSS for glass style UI
st.markdown(
    """
    <style>
    body {
        font-family: -apple-system, BlinkMacSystemFont, "Segoe UI", Roboto, "Helvetica Neue", Arial, sans-serif;
        background-color: #f0f2f6;
        color: #333;
    }
    .container {
        max-width: 800px;
        margin: 0 auto;
        padding: 2rem;
        background: rgba(255, 255, 255, 0.85);
        border-radius: 16px;
        backdrop-filter: blur(10px);
        box-shadow: 0 4px 30px rgba(0, 0, 0, 0.1);
        border: 1px solid rgba(255, 255, 255, 0.3);
    }
    .log-entry {
        padding: 1rem;
        margin-bottom: 1rem;
        background: rgba(255, 255, 255, 0.9);
        border-radius: 10px;
        backdrop-filter: blur(5px);
        box-shadow: 0 2px 10px rgba(0, 0, 0, 0.1);
    }
    .title {
        text-align: center;
        color: #007aff;
    }
    .subtitle {
        text-align: center;
        color: #333;
        margin-bottom: 1.5rem;
    }
    </style>
    """, 
    unsafe_allow_html=True
)

st.markdown('<div class="container">', unsafe_allow_html=True)
st.markdown("<h1 class='title'>Medication Tracker 💊</h1>", unsafe_allow_html=True)
st.markdown("<p class='subtitle'>Track your prescription medications easily and get helpful insights.</p>", unsafe_allow_html=True)

# Form to log medication
st.markdown('<div class="log-form">', unsafe_allow_html=True)
st.markdown("<h2>Log Your Medication</h2>", unsafe_allow_html=True)
with st.form(key='log_medication'):
    medication_name = st.text_input("Medication Name", key='medication_name', placeholder="e.g., Aspirin")
    dosage = st.text_input("Dosage (e.g., 500 mg)", key='dosage', placeholder="e.g., 500 mg")
    time_of_day = st.selectbox("Time of Day", ["Morning", "Afternoon", "Evening", "Night"], key='time_of_day')
    date = st.date_input("Date", datetime.today(), key='date')
    submit_button = st.form_submit_button(label='Log Medication')

    if submit_button and medication_name and dosage:
        add_medication(medication_name, dosage, time_of_day, date)
        st.success(f"Logged: {medication_name} - {dosage} at {time_of_day} on {date}")

st.markdown('</div>', unsafe_allow_html=True)

# Display logged medications
st.markdown('<div class="logged-medications">', unsafe_allow_html=True)
st.markdown("<h2>Logged Medications</h2>", unsafe_allow_html=True)
medications = get_all_medications()
if not medications:
    st.markdown("No medications logged yet.", unsafe_allow_html=True)
else:
    for med in medications:
        st.markdown(f'<div class="log-entry"><strong>{med[1]}</strong><br>Dosage: {med[2]}<br>Time: {med[3]}<br>Date: {med[4]}</div>', unsafe_allow_html=True)

st.markdown('</div>', unsafe_allow_html=True)

# Provide insights and data visualization
st.markdown('<div class="insights">', unsafe_allow_html=True)
st.markdown("<h2>Insights and Data Visualization</h2>", unsafe_allow_html=True)
if medications:
    df = pd.DataFrame(medications, columns=["ID", "Medication", "Dosage", "Time", "Date"])
    
    # Convert Dosage to numeric
    df['Dosage'] = df['Dosage'].str.extract('(\d+)').astype(float)
    
    most_common_med = df['Medication'].mode()[0]
    st.markdown(f"Most frequently taken medication: {most_common_med}", unsafe_allow_html=True)
    
    avg_dosage_time = df.groupby('Time').size().idxmax()
    st.markdown(f"Most common time of day for taking medications: {avg_dosage_time}", unsafe_allow_html=True)
    
    st.markdown("Take your medications consistently at the same time each day to maintain stable drug levels in your body.", unsafe_allow_html=True)
    st.markdown("If you have multiple medications, consult your healthcare provider to ensure there are no adverse interactions.", unsafe_allow_html=True)

    # Visualization
    st.markdown("<h3>Medication Intake Over Time</h3>", unsafe_allow_html=True)
    df['Date'] = pd.to_datetime(df['Date'])
    fig, ax = plt.subplots()
    for key, grp in df.groupby(['Medication']):
        ax = grp.plot(ax=ax, kind='line', x='Date', y='Dosage', label=key)
    date_form = DateFormatter("%Y-%m-%d")
    ax.xaxis.set_major_formatter(date_form)
    plt.xlabel("Date")
    plt.ylabel("Dosage (mg)")
    plt.title("Medication Intake Over Time")
    plt.legend(title='Medication')
    st.pyplot(fig)

    # Data export
    csv = df.to_csv(index=False)
    b64 = base64.b64encode(csv.encode()).decode()  # B64 encoding
    href = f'<a href="data:file/csv;base64,{b64}" download="medications.csv">Download CSV</a>'
    st.markdown(href, unsafe_allow_html=True)

    # AI Prediction: Predicting missed doses
    st.markdown("<h3>Missed Dose Prediction</h3>", unsafe_allow_html=True)
    df['Missed'] = np.random.choice([0, 1], size=len(df))  # Simulate missed doses for training

    # Prepare the data
    X = df[['Dosage', 'Time']].replace({"Morning": 0, "Afternoon": 1, "Evening": 2, "Night": 3})
    y = df['Missed']
    X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.2, random_state=42)

    # Train the model
    model = GradientBoostingClassifier()
    model.fit(X_train, y_train)

    # Predict the probability of missing the next dose
    user_dosage = st.number_input('Enter your dosage for the next dose', value=0)
    user_time = st.selectbox('Select the time of day for the next dose', ["Morning", "Afternoon", "Evening", "Night"])
    user_time_encoded = ["Morning", "Afternoon", "Evening", "Night"].index(user_time)

    if st.button('Predict Missed Dose'):
        user_data = pd.DataFrame({'Dosage': [user_dosage], 'Time': [user_time_encoded]})
        missed_prob = model.predict_proba(user_data)[:, 1][0]
        st.markdown(f"**Probability of missing the next dose: {missed_prob:.2f}**")

    # Medication Interaction Warnings
    st.markdown("<h3>Medication Interaction Warnings</h3>", unsafe_allow_html=True)
    interactions = {
        "Aspirin": ["Warfarin", "Ibuprofen"],
        "Ibuprofen": ["Aspirin", "Warfarin"],
        "Warfarin": ["Aspirin", "Ibuprofen"],
    }

    user_medications = st.multiselect(
        'Select your current medications',
        list(interactions.keys())
    )

    potential_interactions = []
    for med in user_medications:
        for interaction in interactions.get(med, []):
            if interaction in user_medications and interaction not in potential_interactions:
                potential_interactions.append(interaction)

    if potential_interactions:
        st.warning(f"Potential interactions detected: {', '.join(potential_interactions)}")
    else:
        st.success("No interactions detected.")

    # Adaptive Reminders (Simulated)
    st.markdown("<h3>Adaptive Reminders</h3>", unsafe_allow_html=True)
    reminder_times = {
        "Morning": 8,
        "Afternoon": 13,
        "Evening": 18,
        "Night": 22
    }
    
    def adaptive_reminder(user_time, success):
        reminder_times[user_time] = reminder_times.get(user_time, 0) + (1 if success else -1)
        return reminder_times[user_time]

    if st.button('Simulate Reminder'):
        success = np.random.choice([True, False])  # Simulate user response to reminder
        new_time = adaptive_reminder(user_time, success)
        st.markdown(f"Adaptive reminder time for {user_time} is now set to {new_time}:00")

    # Anomaly Detection in Medication Patterns
    st.markdown("<h3>Anomaly Detection</h3>", unsafe_allow_html=True)
    if len(df) > 10:  # Ensure enough data for anomaly detection
        isolation_forest = IsolationForest(contamination=0.1)
        df['Anomaly'] = isolation_forest.fit_predict(df[['Dosage']])
        anomalies = df[df['Anomaly'] == -1]
        
        if not anomalies.empty:
            st.markdown("### Anomalies Detected:")
            st.dataframe(anomalies[['Medication', 'Dosage', 'Time', 'Date']])
        else:
            st.markdown("No anomalies detected in your medication patterns.")

st.markdown('</div>', unsafe_allow_html=True)

# Footer
st.markdown("<hr>", unsafe_allow_html=True)
st.markdown("<p><strong>Disclaimer:</strong> This app is for tracking purposes only. Always consult with your healthcare provider for medical advice.</p>", unsafe_allow_html=True)
st.markdown('</div>', unsafe_allow_html=True)