Cutting-Edge AI to Manage Stress: Real-Time Tailored Solutions

Spanish Version

Artificial intelligence has proven to be a powerful tool for addressing complex problems across various disciplines, including psychology. In this post, we will explore how real data related to anxiety and depression can be used to create a recommendation system that helps individuals manage their stress in a personalized way. By analyzing key factors in mental health and employing advanced algorithms, this project aims to demonstrate how technology can be an ally in emotional well-being.

Mental health is a priority in the society we live in. With this system, we aim to identify and classify profiles with high levels of stress, anxiety, or depression based on similarities among them. Once these profiles are identified, the goal is to offer personalized recommendations tailored to the individual needs of each person, as well as provide practical strategies to help improve their quality of life.

The system is designed in three fundamental stages that ensure its effectiveness and precision. First, data preprocessing is carried out, involving the preparation and cleaning of information to guarantee its quality and usability. Next, user clustering is implemented using the K-Means algorithm, which allows for the identification and classification of profiles based on patterns and similarities in the data. Finally, the Decision Trees method is used to generate personalized recommendations that adapt to the specific needs of each identified profile, optimizing results and promoting solutions tailored to each case.

K-Means profile identification

For the classification and grouping into different profiles, K-Means clustering algorithm is applied to classify users based on their levels of stress, depression, and anxiety: 

• Configuration: Five groups or "clusters" have been defined to identify patterns within the data:

    kmeans = KMeans(n_clusters=5)

    df['Perfil'] = kmeans.fit_predict(

        df[['Anxiety_Score',

            'Depression_Score',

            'Stress_Level']])

With this graph, we can visually represent the groupings for each identified profile. 

• Group Analysis: The values are analyzed to precisely understand each generated profile. To achieve this, we use various graphs that allow us to better comprehend the data, define, and personalize the profiles further.

from mpl_toolkits.mplot3d import Axes3D

    fig = plt.figure(figsize=(10, 7))

    ax = fig.add_subplot(111, projection='3d')

    for cluster in df['Perfil'].unique():

        cluster_data = df[df['Perfil'] == cluster]

        ax.scatter(cluster_data['Stress_Level'],

                   cluster_data['Anxiety_Score'],

                   cluster_data['Depression_Score'],

                   label=f'Perfil {cluster}')

    plt.show()

    profiles = df.groupby('Profile')[['Stress_Level',

                                    'Anxiety_Score',

                                    'Depression_Score']].mean()

    

    profiles.plot(kind='bar', figsize=(10, 6), cmap='viridis', edgecolor='black')

    plt.xlabel('Profile')

    plt.show()

• Associated recommendations: Each profile has a personalized strategy designed to improve the user's well-being.

Customized prediction with Decision Tree

In recommender system development, one of the key challenges lies in identifying meaningful patterns within the data that enable tailored solutions for individual users. To address this, a decision tree model has been implemented, leveraging relevant features and utilizing the profile variable created through K-means clustering as the recommendation target.

This approach facilitates the detection of patterns based on user characteristics, enabling personalized recommendations aligned with their unique needs. The methodology prioritizes a balance between interpretability and effectiveness, resulting in a recommendation system that is both accurate and easy to understand.

The accompanying graph offers a simplified visualization of the decision tree's training process, illustrating the paths it follows to arrive at a conclusion. It emphasizes how the tree identifies the most suitable profile for each user, based on the analyzed values.

This system provides a comprehensive tool to promote emotional and psychological well-being, offering substantial benefits in three key areas:

1. Early detection: by analyzing behavioral patterns and emotional states, the system enables early identification of individuals experiencing elevated levels of stress, anxiety or depression. This supports timely interventions, helping to manage problems before they escalate and ensuring rapid and effective action.
2. Personalized recommendations: the system provides customized strategies designed to meet the unique needs of each user. From simple relaxation techniques to practical tips for organizing daily life, these recommendations are designed to be accessible and effective, promoting noticeable improvements in overall well-being.
3. User-friendly design: designed with inclusion in mind, the system is accessible to everyone, regardless of their technical knowledge. Its intuitive interface ensures a smooth and enjoyable experience, making it easy for users to access valuable resources and tools to improve mental and emotional health.

Fundamentally, this system becomes a vital ally on the path to individual wellness, offering a proactive, personalized and accessible solution.

Now it's your turn! How well do you understand your levels of stress, anxiety, or depression? Discover the answers with this recommendation system and receive personalized strategies that could transform your daily routine and enhance your well-being.

Ready to give it a try?

If you're curious about how it works, how to implement this system, or how to customize it to your needs, I'm here to offer more ideas and technical insights to help you along the way!

Let me know if you'd like to tweak it further!

Strategy Recommender Based on Your Stress Levels

How old are you? (Age): What is your gender? (1: Female, 0: Male): What is your educational level? (0: No studies, 1: High School/Vocational Training, 2: Bachelor's, 3: Master's, 4: Doctorate): What is your employment status? (0: Unemployed, 1: Employed, 2: Student, 3: Retired): How many hours do you sleep per day? (0-12): How many hours of physical activity do you do per week? (0-10): Do you have a family history of mental illness? (0: No, 1: Yes): Do you have any chronic illnesses? (0: No, 1: Yes): Do you take medication regularly? (0: No, 1: Occasionally, 2: Frequently, 3: Always): Have you ever undergone therapy? (0: No, 1: Yes): Do you practice meditation? (0: No, 1: Yes): Do you consume alcohol or tobacco? (0: No, 1: Occasionally, 2: Frequently, 3: Always): What is your level of financial stress? (1-10): Do you experience work-related stress? (1-10): What is your level of self-esteem? (1-10): How would you rate your life satisfaction? (1-10): Do you frequently feel lonely? (Loneliness level, 1-10):