About MindScan

How It Works

An XGBoost classifier trained on 1,200 teen mental health records to predict depression risk.

Dataset

1,200 teen records covering social media habits, sleep, academics, stress, anxiety, and addiction levels.

Preprocessing

Label encoding for categorical features + SMOTE oversampling to handle class imbalance (~2.6% positive cases).

Model

XGBoost — a gradient boosted tree ensemble known for speed and accuracy on tabular data.

Output

0 = Low Depression Risk  |  1 = Depression Risk Detected, plus a confidence score.

Input Features

#FeatureTypeRange / Values
1AgeNumeric10 – 25
2GenderCategoricalMale / Female
3Daily Social Media HoursNumeric0 – 24 hrs
4Platform UsageCategoricalInstagram / TikTok / Both
5Sleep HoursNumeric0 – 12 hrs
6Screen Time Before SleepNumeric0 – 10 hrs
7Academic PerformanceNumericGPA 0.0 – 4.0
8Physical ActivityNumeric0 – 10 hrs/day
9Social Interaction LevelCategoricalLow / Medium / High
10Stress LevelNumeric1 – 10
11Anxiety LevelNumeric1 – 10
12Addiction LevelNumeric1 – 10
Developer

Shubham Kumar

ML Engineer & Full Stack Developer passionate about using AI for social good.

Go to Assessment